Low-Dose Quantitative Myocardial Blood Flow Imaging Using 15O-Water and PET Without Attenuation Correction

Abstract

Misalignment between PET and low-dose CT (LD-CT) can cause severe artifacts in cardiac PET/CT because of attenuation-correction errors, even when using slow or cine LD-CT. Myocardial blood flow (MBF), as measured by (15)O-water, can be determined from the rate of (15)O-water washout from myocardial tissue, which is independent of tissue attenuation. The purpose of the present study was to assess the accuracy of these MBF measurements in the absence of attenuation correction. Twenty-five patients referred for evaluation of myocardial perfusion underwent 6-min rest and adenosine stress PET scans after the administration of 370 MBq of (15)O-water; both scans were followed by slow LD-CT. Data were acquired on a PET/CT scanner and reconstructed by a 3-dimensional row-action maximum likelihood algorithm both with (CTAC) and without (NAC) attenuation correction. An ascending aorta volume of interest was used as input function. MBF and coronary flow reserve (CFR) were calculated for 17 myocardial segments using nonlinear regression of the standard single-tissue-compartment model with corrections for left and right ventricular spillover and perfusable tissue fraction. High correlation (r(2) = 0.99 and 0.97, with slopes of 0.96 and 0.91 for rest and stress, respectively) and excellent agreement (intraclass correlation coefficient [ICC], 1.00 and 0.98) between NAC- and CTAC-based MBF values were found. Absolute rest and stress MBF values were 3% and 8%, respectively, lower for NAC scans. The correlation coefficient between all NAC and CTAC CFR values was 0.95 (ICC, 0.95; slope, 0.92) and 0.97 (ICC, 0.99; slope, 1.01) when only CFR values below 2 were considered. Deviations between CTAC and NAC values were smallest for basal segments and increased toward the apex. MBF and CFR can be measured accurately using (15)O-water and PET without correcting for attenuation, reducing the effective dose to the patient to 0.8 mSv for a complete rest-stress protocol. This dose is an order of magnitude lower than typical values for (82)Rb, (99m)Tc-methoxyisobutylisonitrile, or CT perfusion scans.